Return to “Polls”

What do you prefer to think of workers as?

Humans - everything is humans
Total votes: 13 (17%)
Biological sentients - species mix and match
Total votes: 25 (32%)
Artificial intelligences - just programs running a ship
Total votes: 11 (14%)
Robots - programs with a mechanical body
Total votes: 11 (14%)
Androids - robots with some biological functions
Total votes: 10 (13%)
Clones of the NPC - exactly what it says on the tin
Total votes: 2 (3%)
Quasi-clones of the NPC - clones, but with slightly varying personalities
Total votes: 2 (3%)
Other - detail in a reply
Total votes: 4 (5%)
Total votes: 78
Post

Re: What should workers be?

#16
I come back to Josh's statement that workers will have simplified AI for performance reasons.

To me, that needs a diegetic explanation of some kind, and that explanation should suggest additional fun gameplay features that follow from its form... but that form could successfully be any number of things.

In addition, I'm thinking most (human) players of LT will prefer to see their character -- even if they never "see" an avatar of that character -- as a person. And since Josh has said (as I recollect it) that he wants NPCs (now meaning executives) and the human player to have as much parity as possible, then by extension that means executive NPCs would need to be notional people as well as the player's character.

So the remaining question -- for me, anyway :) -- is about workers. Why is their processing simpler compared to real people? What form could they have that explains this that's plausible for LT without being too complicated?

Pure programs running ships is a reasonable suggestion. Not super-exciting as it doesn't really connect to other LT systems, but workable.

The clones-of-executives idea is more interesting as it connects to the idea of factions (with workers forming a kind of futuristic "family" with their executive-parent). But a perfect or minimally varied clone would be a person, and so wouldn't explain why workers have lower decision-making capabilities.

So in this case, I like the idea of workers being different subsets of their originating executive's mind. Maybe they inherit two out of the executive's personality characteristics. This lets them express different behaviors in different circumstances, and allows them to be optimized for different kinds of work in an organization. It also explains why they're less competent than an executive. It doesn't lend itself to promotion to executive-hood, but then the ideas of workers as ship-AI or robots don't, either.

Are there other forms that would offer better ways of satisfying these design needs? Or are there other constraints that workers need to satisfy that I'm not seeing?
Post

Re: What should workers be?

#17
Flatfingers wrote:So the remaining question -- for me, anyway :) -- is about workers. Why is their processing simpler compared to real people? What form could they have that explains this that's plausible for LT without being too complicated?
If that's the only remaining question, then great, because my proposal addresses that. The difference between workers and executives, in my understanding, is that workers are strictly goal-solving - they attempt to solve goals given to them - while executives can be both goal-solving and goal-setting, where they consider different options, estimate the utility of each and compare them against one another. Executives think of goals and then assign them to workers (or other executives) to solve.

How is this explained in lore? Easily - all you need to do is imagine that workers are completely loyal to their commanding executives, and so obey any order given to them. Workers, as the digital "children" of executives, would obey their executives out of a mixture of respect, fear (they can be terminated at any moment) and a sense of shared identity or belonging. While it might be cool and still more plausible to see "children" betray their "fathers" once in a while, this isn't computationally feasible and in any case would probably happen seldom enough that we can omit that from the game without any huge loss to gameplay or plausibility.

Workers are not "simple", and are capable of solving any goal just as well as an executive. They can manage hundreds of other workers if necessary. And to nip the "low-LOD NPCs shouldn't be capable of this kind of stuff" objection in the bud before someone raises it again: the whole point that low-LOD agents were introduced was to cut down on CPU overhead so that we can get decent framerates without having to restrict system populations to a few dozen guys. Now, as you progress and if you ever want to establish your own empire, you will need people who can manage dozens or hundreds of other agents, or who can handle your finances or your fleets, etc. So you have two choices here: hiring executives or hiring workers. Now, assuming workers are capable of solving high-level goals as I propose, which organisation system do you think would be less computationally demanding: one composed almost entirely of executives, or one composed almost entirely of workers? I make the same point here.

And as for the objection that allowing everyone in an organisational system to be a worker will obviate inter-executive cooperation - no it won't. That is addressed here.
Post

Re: What should workers be?

#19
ThymineC wrote:
5anitybane wrote:
Grumblesaur wrote:I would say a mix of wholly biological sentients, androids and cyborgs, and full-blown robots.
Exactly my selected options as well. Makes sense to have biologicals for those unable to afford purchasing more expensive robotics, then the middle-ground of androids/cyborgs for those slightly better off, and full-blown, highly-efficient robots for the wealthy entrepreneurs.
So in your view, being more mechanical is considered to be the more desirable option within the LT universe?
I think he's saying that robots or mechanically-boosted sentients are more durable and capable, therefore more expensive, which is why wealthier people would have access to them.

But here's the idea:
  • Biological Sentients:
    • Pros: Able to fill in gaps in instructions, easily retrained, very common
    • Cons: Unable to work in extreme environments, require wages and/or rations
  • Cyborgs / Mechanically-Boosted Sentients:
    • Pros: Able to fill in gaps in instructions, easily retrained, can handle more physical stress
    • Cons: Unable to work in extreme environments, require wages and/or rations, uncommon
  • Androids / Sentient Robots:
    • Pros: Very durable, able to fill in gaps in instructions, able to work in extreme environments
    • Cons: Expensive, require charging points and maintenance infrastructure
  • Non-Sentient Robots / Machines:
    • Pros: Very durable, able to work in extreme environments, very fast operation
    • Cons: Unable to fill in gaps in instructions, require other machines to complete tasks, require maintenance infrastructure
Shameless Self-Promotion 0/ magenta 0/ Forum Rules & Game FAQ
Post

Re: What should workers be?

#20
Grumblesaur wrote:I think he's saying that robots or mechanically-boosted sentients are more durable and capable, therefore more expensive, which is why wealthier people would have access to them.
Yeah. I said it semi-rhetorically. I think it's awesome that he thinks that way. :thumbup: I like your list.
Post

Re: What should workers be?

#21
Yeah, that was essentially what I said, though there was also the 'mushy' factor of sentient beings and their ability for free-thought, which is both a pro and con if there are alternatives. My explanation went into more detail as to why I thought this, :).

I like the list, more... succinct than I wrote, hahaha.
Post

Re: What should workers be?

#22
ThymineC wrote:The difference between workers and executives, in my understanding, is that workers are strictly goal-solving - they attempt to solve goals given to them - while executives can be both goal-solving and goal-setting, where they consider different options, estimate the utility of each and compare them against one another.
...
Workers are not "simple", and are capable of solving any goal just as well as an executive. They can manage hundreds of other workers if necessary.
Hmm.

Josh did say, on Nov. 21, 2013:
Josh Parnell wrote:I will give some NPCs more "neurons" and such so that they will make deeper and more accurate decisions. This should lead to them ultimately being able to control a lot of "less smart" workers by hiring. But to construct two different types of AI - one "stupid" and one "smart" is, IMO, counterproductive to the soul of this game :)

The current AI should be scalable enough that a similar effect can be achieved even while individual actors maintain the ability to think for themselves :thumbup:
But more recently, on April 6, 2014 was the big worker/player breakthrough:
Josh Parnell wrote:[W]e will treat living entities as being split into two categories: high level-of-detail and low level-of-detail. Low-LOD entities are those that will be performing labor, while high-LOD entities are those that are performing thought and management.
...
By treating personnel at two different levels of detail, it almost goes without saying that we relieve ourselves of a lot of the computational cost of AI.
...
Under this model, you can see that PC-NPC symmetry is fully-preserved. Players are all equal in their capacities. This is because we have moved the asymmetry into the worker-player split. Players are not equivalent to workers in their capacities.
This may have evolved since then. But I read it as workers being different from executives ("players") not just in degree but in kind -- and necessarily so in order to gain the necessary performance optimization.
ThymineC wrote:And to nip the "low-LOD NPCs shouldn't be capable of this kind of stuff" objection in the bud before someone raises it again: the whole point that low-LOD agents were introduced was to cut down on CPU overhead so that we can get decent framerates without having to restrict system populations to a few dozen guys.
Heh. :) I guess I'm raising it again just the same, as cutting down on CPU overhead does suggest to me that workers have a functionally reduced capacity for interesting behaviors. AI running at a lower level of detail means not merely different behaviors, but less capacity for more generally responsive behaviors -- by design.

Again, though, I don't think that by itself is enough to dictate the form of workers. Less interesting == less important to me, so while I like 'bots less than reduced-clones, I expect I'd be OK either way.

I'll be spending my time poking the executives to see how they jump. :)
Post

Re: What should workers be?

#23
Flatfingers wrote: I'll be spending my time poking the executives to see how they jump. :)
And when they start poking back to see how you jump?
Image
Challenging your assumptions is good for your health, good for your business, and good for your future. Stay skeptical but never undervalue the importance of a new and unfamiliar perspective.
Imagination Fertilizer
Beauty may not save the world, but it's the only thing that can
Post

Re: What should workers be?

#24
Hyperion wrote:
Flatfingers wrote: I'll be spending my time poking the executives to see how they jump. :)
And when they start poking back to see how you jump?
Depends on which "I" we're talking about.

I meant "I-the-human-player-of-a-game," in which case, Skynet Bad.

But I'd love to see NPCs test "I-the-game-character" to observe and exploit patterns. I don't expect that, but it would be interesting.
Post

Re: What should workers be?

#25
Flatfingers wrote: Josh did say, on Nov. 21, 2013:
[...]

But more recently, on April 6, 2014 was the big worker/player breakthrough:
[...]
Josh said in IRC a while back that worker NPCs should at least be able to perform fleet leadership. That sounds like management and places them at least at the level of being capable of operational planning.
Flatfingers wrote:By treating personnel at two different levels of detail, it almost goes without saying that we relieve ourselves of a lot of the computational cost of AI.
Indeed.
Flatfingers wrote: Heh. :) I guess I'm raising it again just the same, as cutting down on CPU overhead does suggest to me that workers have a functionally reduced capacity for interesting behaviors. AI running at a lower level of detail means not merely different behaviors, but less capacity for more generally responsive behaviors -- by design.
And they do have less capacity for certain kinds of behaviours - for one, we eliminate their ability to set their own goals. All they do is take goals from another agent (their commanding executive) and figure out how to solve those goals. Workers have the ability to solve those goals perhaps just as well as executives, however.

And why? Because you (and other execs) are going to need people to manage your assets no matter what - you will need to hire NPCs that are capable of solving goals. By designing worker NPCs to be as capable as executives at solving goals and then positing no additional "thinking" that they do beyond that, it allows for the most efficient, more CPU-efficient implementation possible, that I can see. The player and other execs will need to hire NPCs that are capable of solving goals, so why not let them hire NPCs that are capable of solving goals and do nothing beyond that? The alternative that you seem to be proposing - to keep workers dumb - would necessitate that players and execs hire other execs to do management and such for them. And that would mean that people within your workforce are not just trying to solve goals but trying to create their own goals as well - which might be cool, but it's far more computationally expensive.

In my proposal, worker NPCs are only as smart as they need to be - or alternatively, as dumb as they can afford to be - and so the only CPU that is allocated to them is CPU that necessarily has to be allocated to them, and would still be allocated to them along with a lot of unnecessary CPU as well if they were executives instead. If you composed your entire workforce out of worker NPCs, as I propose is possible, then you get the most CPU-efficient setup.

Online Now

Users browsing this forum: No registered users and 12 guests

cron